The Bartlett
B-Pro Show 2022

Explore

About the show

0 other visitors

Mute Unmute Exit

unit-code

Project sections

1 / 1

Informatisation City

Project details

Programme

Cluster RC11

Award

B-Pro Urban Design Silver Medal

The 21st century is a new information age. The vast amount of information has completely changed the ways cities operate and are organised. Information challenges existing approaches to urban design and offers the possibility of making urban design smarter, more flexible and more efficient. We need to make information sing.

To do this, this project uses information and machine learning to design cities, and built a new urban design framework consisting of the ICI system and the GAN2City generator. The ICI system is a framework for merging information from different views of the city in response to specific problems, and is used to help us understand the city with information. The GAN2City generator receives the processed information from the ICI system, understands and learns the features from this information, discovers the 'hidden layers' behind this information and generates a new city from it. In the scenario of an enhanced environment, we have used this new framework to create a new La Défense.

Students

Background and Concept

Core Concepts

The background to this project and two core concepts: cognitive cities through information and designing cities with machine learning and information.

Urban Design in the New Information Age

The design methods that rely heavily on traditional urban cognitive methods and urban theories and models will no longer be suitable for the new information age.

City Cognition Through Information

In the new information age we need a more formal, objective and broader approach to urban cognition; information.

Urban Design With ML and Information

In the face of the increasing complexities of cities, a new approach to urban design is needed that can integrate the increasing complexity of cities, while being real-time and resilient; it is machine learning.

ICI System

In this section, we will describe in further detail how the ICI system works.

ICI System for La Défense

For La Défense we constructed four epistemic planes: spatial structure, ICT networks, functioning of key aspects and the plane of agents and forces.

Data Collection, Processiong and Response

This video shows how we collect large amounts of data and process them into semantic information that can describe La Défense according to a specific scenario.

Data Clustering

For the four scenario-based city's epistemic planes we obtained, we used clustering to extract the features and merged them in a way that did not reduce the dimensionality of the information for subsequent operations.

Output of the ICI System

Every plot and every space can be described by four dimensions of information.

Results of Urban Spatial Layout Optimisation

Based on the ICI results, the spatial layout of La Défense was optimised using NSGA-II to obtain the most favourable spatial type and arrangement for the current scenario.

The Training Process

In this section, we further discuss the working principles and processes of GAN2City, a machine learning framework for urban design.

Framework of GAN2CITY Generator

The GAN2CITY Generator consists of four main steps: GAN2PLAN, GAN2BOX, GAN2VIEW and 3D REBUULDING.

GAN2PLAN

GAN2Plan is an artificial intelligence model that can transform plot shapes into plot layouts in urban planning.

GAN2BOX

Gan2Box generates building physiognomy based on the previous step of planes through unpaired learning, and then outputs a series of axonometric drawings.

GAN2VIEW

GAN2View is an artificial intelligence model that can add façades details to the buildings boxes.

Iterations of La Défense

This video shows the evolution of the La Défense over 20 generations, including plot grade, building form and building façades.

Building Structure and Design Output

This section will show a new, completely information-generated city and more specific internal building spaces and structures.

The Output of Building Structures

We select and generate appropriate and detailed building interiors and structures based on the information from the GAN2City Generator.

Design Output

Design Process of Informatisation City

View Model

The Building in Plot 64

This project represents our vision of the city of the future. We want it to be more scientific, more flexible, more adaptable, more efficient and smarter.

Share on Facebook, LinkedIn or Twitter

Index of Works

The Bartlett
B-Pro Show 2022

27 September – 7 October

Explore